""" collect.py Functions for collecting data for insertion into the databse or pulling data from the database for analysis """ from datetime import datetime, timedelta from time import sleep from sqlalchemy import func from sqlalchemy.exc import IntegrityError from data.database import Base, dal import util.drawing as drawutil import util.environment as envutil import util.scrape as scrapeutil project_variables = envutil.load_environment_variables() def get_record_count(table_name: str) -> int | None: """get the most recent draw record from the specified table""" game_dict = drawutil.check_table_name(table_name=table_name) if game_dict: dal.connect() session = dal.Session() target_table = Base.metadata.tables.get(game_dict['db_table_name']) return session.query(func.count(target_table.columns.draw_date)).first()[0] else: raise ValueError('An invalid table name was specified.') def get_recent_draw_date(table_name: str) -> datetime | None: """get the most recent draw date from the specified table""" game_dict = drawutil.check_table_name(table_name=table_name) if game_dict: dal.connect() session = dal.Session() target_table = Base.metadata.tables.get(game_dict['db_table_name']) return session.query(func.max(target_table.columns.draw_date)).first()[0] else: raise ValueError('An invalid table name was specified.') def get_missing_records(table_name: str) -> int: """collect and insert missing records, returning a count of insertions""" # if the table_name is valid game_dict = drawutil.check_table_name(table_name=table_name) if game_dict: insert_count = 0 # connect to the database and establish a session dal.connect() session = dal.Session() date_format = project_variables['LP_DATE_PASS_FORMAT'] target_table = Base.metadata.tables.get(game_dict['db_table_name']) record_count = get_record_count(table_name=table_name) target_game = game_dict['game_name'] end_year = datetime.today().year + 1 recent_draw_date = (get_recent_draw_date(table_name=table_name) + timedelta(days=1)).strftime(date_format) following_draw_date = datetime.strptime(drawutil.get_next_draw_date(game=target_game, from_date=recent_draw_date), date_format) start_year = game_dict['inception_year'] if record_count == 0 else following_draw_date.year check_date = datetime.strptime(drawutil.get_previous_draw_date(game=target_game, from_date= drawutil.get_next_draw_date(game=target_game)), date_format).date() if session.query(target_table).filter_by(draw_date=check_date).first() is None: # loop over the range of years and insert missing records for year in range(start_year, end_year): scrape_data = scrapeutil.scrape_game_history(game=target_game, year=year) # convert and insert records, discarding duplicate dates for draw_date, draw_info in scrape_data.items(): record = game_dict['record_type'](draw_date=draw_date, draw_result=draw_info) try: session.add(record) session.commit() insert_count += 1 except IntegrityError: session.rollback() if year != end_year - 1: sleep(30) # sleep between calls to the site for data to avoid being blocked return insert_count else: raise ValueError('An invalid table name was specified.')